Vannevar Bush’s 1945 Vision and the AI-Internet World It Inspired
In July 1945, as the Second World War was ending, American scientist and engineer Vannevar Bush published an essay in The Atlantic Monthly that would quietly shape the future of information technology. Titled “As We May Think,” it wasn’t a military triumphalist piece, even though Bush had overseen the U.S. Office of Scientific Research and Development, coordinating the efforts of thousands of scientists on radar, weapons, and medical innovations. Instead, Bush called for something radical in its peacefulness: that scientists turn from building destructive machines to creating tools that amplify the h
uman mind.
Bush saw the intellectual crisis of his time with startling clarity. Scientific knowledge was expanding faster than anyone could keep up with, and the systems for storing and retrieving it—libraries, journals, filing cabinets—were hopelessly outdated. If something didn’t change, he warned, science could choke on its own success.
Seventy-five years later, as we navigate an era of search engines, Wikipedia, and generative AI, Bush’s essay reads less like quaint speculation and more like a foundational blueprint. His most famous concept, the memex, directly inspired hypertext, personal computing, and, ultimately, the web. But just as striking are the ways in which his hopes—and his anxieties—map onto our own AI-driven world.
This is the story of how Bush imagined the future, how we got here, and whether we’re finally solving the knowledge crisis he feared—or just building a shinier version of it.
The Crisis of Knowledge Overload
When Bush wrote As We May Think, the problem wasn’t that humanity lacked knowledge—it was that knowledge was becoming unusable.
The volume of scientific research had exploded during the war, and Bush saw a “growing mountain of research” burying crucial insights. Mendel’s laws of genetics, he reminded readers, had been published in 1866 but went unnoticed for 35 years because no one had stumbled across them in time. “This sort of catastrophe is undoubtedly being repeated all about us,” he warned.
The methods of information retrieval were medieval in pace: alphabetical indexes, subject hierarchies, printed bibliographies. Finding a key piece of research meant following a rigid paper trail through stacks of journals—an inherently linear and slow process. Bush compared this to “using a stone adze in the hands of a cabinetmaker.”
If knowledge production continued to accelerate while retrieval methods stayed static, progress could stall. Science risked a soft collapse, not by losing knowledge but by making it effectively invisible.
Bush’s solution? Build machines to extend the powers of the human mind, just as microscopes and engines had extended our physical abilities.
The Memex: A Desk-Sized Internet Before the Internet
Bush’s answer to information overload was the memex, a hypothetical device he described in loving detail.
The memex would look like a desk with screens, levers, and buttons. Inside, it stored vast amounts of information on microfilm—entire libraries compressed into a few drawers. But the real innovation wasn’t storage. It was linking.
Bush imagined users creating “associative trails,” connecting documents the way the human mind jumps from one idea to another. Instead of relying on alphabetical or numerical indexing, a user could create a personal web of knowledge, weaving together books, articles, and notes into meaningful paths.
He even anticipated social sharing: one could copy and send these trails to colleagues, who could merge them into their own. In other words, Bush described hypertext, bookmarks, and collaborative knowledge curation decades before the web existed.
“Selection by association, rather than by indexing, may yet be mechanized,” Bush wrote. This single sentence contains the seed of almost every major information technology advance since.
Other Predictions: Computers, Voice Recognition, and Beyond
The memex is Bush’s most famous idea, but As We May Think is full of prescient details:
Voice-controlled typewriters: Bush speculated that scientists would dictate directly to machines. Today, we call this speech-to-text.
Miniature wearable cameras: He imagined scientists wearing walnut-sized cameras to record observations hands-free—a clear ancestor to GoPros and AR glasses.
Rapid computation: Bush foresaw “electrical arithmetic machines” capable of solving complex equations at unprecedented speeds. Within a decade, early computers like ENIAC would prove him right.
Trail blazers: He predicted new professions dedicated to curating and sharing knowledge pathways—something we can recognize in modern bloggers, Wikipedia editors, and even data journalists.
Neural interfaces: In a wild flourish, Bush speculated that someday we might bypass keyboards entirely, feeding information directly into the brain. In 1945, this sounded like science fiction; today, brain-computer interfaces are experimental reality.
Did We Avoid the Collapse He Feared?
The short answer: yes, but only by transforming the very systems Bush criticized.
Postwar Reforms and Early Mechanization
The first response to Bush’s challenge was institutional rather than technological. Scientific communities adopted better indexing (like the Science Citation Index in 1960), and collaborative “big science” projects—NASA, CERN—reduced duplication by sharing data across teams.
Microfilm libraries and early computing machines, which Bush partly inspired, also provided temporary relief by compressing and automating parts of the information pipeline.
The Web as Memex Realized
But the real breakthrough was hypertext and the internet. Douglas Engelbart, who invented the computer mouse and worked on early hypertext systems, cited Bush as a direct influence. Ted Nelson, who coined the term “hypertext,” called Bush’s essay “one of the most influential pieces of writing in the history of information science.”
The World Wide Web—an open, hyperlinked, user-generated information system—is effectively a global memex.
Did We Solve the Problem or Just Defer It?
We avoided Bush’s 1945 bottleneck, but the underlying tension hasn’t gone away. Instead of drowning in journals, we now drown in data, misinformation, and algorithmically amplified noise.
The question Bush raised—how can humans keep up with the knowledge they generate?—is still alive, only now the problem isn’t access but filtering truth from noise.
Bush’s Vision and the AI Era
This is where things get interesting: Bush imagined machines that amplified human intellect, not ones that thought for us. His philosophy was deeply human-centric.
AI as the Ultimate Memex?
Modern AI systems like ChatGPT can already do something Bush dreamed of: thread through massive stores of information and surface relevant associations instantly. In a sense, an AI assistant is a memex with predictive capabilities, dynamically building trails rather than relying solely on pre-set ones.
But Bush would likely have been cautious. His memex preserved human agency—the user decided what was relevant, built the trails, and interpreted the results. In contrast, AI models make many of those decisions invisibly, shaping knowledge access through statistical prediction rather than human judgment.
Trail Blazers or Black Boxes?
Bush imagined a future where “trail blazers” became respected professionals, curating meaningful paths through knowledge. Today, that role is partly played by journalists, scientists, and curators—but increasingly, it’s algorithms choosing what billions of people see.
This is a profound shift: we’ve replaced Bush’s ideal of transparent human-curated trails with opaque machine learning systems. The result is sometimes efficient, sometimes manipulative, and often both.
The Optimism and Its Limits
Bush was relentlessly optimistic. He believed that by mechanizing memory, humans could focus on creative and intuitive thought, pushing civilization forward. He envisioned science not just as a tool for survival but for wisdom.
Yet Bush underestimated the social, political, and ethical challenges of knowledge systems. He didn’t foresee how economic incentives might turn information tools into advertising platforms, or how disinformation could weaponize the very systems meant to democratize knowledge.
His essay reads like a love letter to the scientific method and a belief in rational progress—a worldview that feels almost naïve in an age of deepfakes and algorithmic polarization.
Why Bush Still Matters
Despite these differences, Bush’s core insight remains as urgent as ever: humanity’s intellectual progress depends on how well we design our tools for thought.
His emphasis on associative trails is influencing a new wave of personal knowledge management (PKM) tools like Roam Research, Obsidian, and Notion.
His call for “trail blazers” resonates with the movement toward open science and knowledge curation.
His insistence that machines should augment rather than replace human creativity is at the heart of today’s debates over AI alignment.
Bush wasn’t designing gadgets; he was designing a philosophy of knowledge. And that philosophy still has unfinished business.
Conclusion: As We Still May Think
Bush ended his essay with a hopeful vision: that mechanized knowledge tools would allow humans to forget the trivial, remember the essential, and grow wiser as a species. “Man’s spirit should be elevated,” he wrote, “if he can better review his shady past and analyze more completely and objectively his present problems.”
In 2025, we stand on the edge of technologies Bush could barely imagine. We’ve built global memexes, and now we’re building machines that can generate knowledge on demand. Whether these tools lead us to wisdom—or simply deeper confusion—depends on whether we heed Bush’s deeper message:
Technology alone doesn’t make us wiser. But technology designed to honor the way we think might.